key: cord-1013960-3jodvhmj
authors: Ojalehto, Heidi J.; Abramowitz, Jonathan S.; Hellberg, Samantha N.; Butcher, Megan W.; Buchholz, Jennifer L.
title: Predicting COVID-19-Related Anxiety: The Role of Obsessive-Compulsive Symptom Dimensions, Anxiety Sensitivity, and Body Vigilance
date: 2021-07-29
journal: J Anxiety Disord
DOI: 10.1016/j.janxdis.2021.102460
sha: f2989185d3470830ea3149fd407d1fb74cca7b8e
doc_id: 1013960
cord_uid: 3jodvhmj

The novel coronavirus disease (COVID-19), first detected in December of 2019 and declared a global pandemic in March of 2020, continues to pose a serious threat to public health and safety worldwide. Many individuals report anxiety in response to this threat, and at high levels, such anxiety can result in adverse mental health outcomes and maladaptive behavioral responses that have consequences for the health of communities more broadly. Predictors of excessive anxiety in response to COVID-19 are understudied. Accordingly, the present study examined psychological factors that predict more intense COVID-19-related anxiety. 438 community members completed measures assessing COVID-19-related anxiety as well as psychological variables hypothesized to predict anxious responding to the threat of COVID-19. As expected, obsessive-compulsive symptoms related to contamination, the fear of arousal-related body sensations (i.e., anxiety sensitivity), and body vigilance each predicted more severe anxiety related to the pandemic. Obsessive-compulsive symptoms related to responsibility for causing harm also emerged as a predictor. Study limitations and implications are discussed.

Anxiety and fear are common experiences during disease outbreaks, such as the COVID-19 pandemic (e.g., Rubin et al., 2009; Wheaton et al., 2012; Cheng et al., 2004) . Approximately 30% of the general population has reported significant anxiety symptoms since the COVID-19 outbreak in early 2020 (Salari et al., 2020) . Moreover, the intensity of one's anxiety appears to be a significant predictor of dysfunctional behavior amidst the threat of illness, such as frequent hospital visits for minor symptoms and panic buying (e.g., stockpiling toilet paper; Taylor, 2019). Research has also found that greater levels of anxiety about COVID-19 are associated with increased depression and suicidal ideation . Given the importance of anxiety in predicting psychological and behavioral responses to pandemics, it is critical to understand factors that may predict the severity of anxiety in this context. Whereas the majority of research to date has focused on the mental health impact of the COVID-19 pandemic (e.g., Xiong et al., 2020; Serafini et al., 2020) , less attention has been paid to identifying predictors of COVID-19-related anxiety. Prior work has demonstrated that individuals with anxiety-related disorders experience greater COVID-related distress, indexing higher scores on measures of COVID stress than those with mood disorders and non-clinical individuals (e.g., Asmundson et al., 2020) . Further, individuals with generalized anxiety disorder, panic disorder, and obsessive-compulsive disorder (OCD) report greater stress and phobic reactions to COVID-19 compared to individuals with social anxiety and specific phobia (Ardestani et al., 2021; Khosravani et al., 2021b) . Studies have also demonstrated links between health anxiety, defined as the preoccupation with the fear that one has or may contract a serious medical illness (Hedman & Axelsson, 2017) , and COVID-19-related anxiety (e.g., Mertens et al., 2020; Jungmann & Whithoft, 2020; Khosravani et al., 2021b) . However, excessive COVID-19related anxiety may also be a unique variant of health anxiety, in which the fear is specific to contracting COVID-19 rather than illness more generally. Notably, previous research has demonstrated that there is some, but not complete, overlap in psychological predictors implicated in anxiety specific to disease outbreaks and more generalized health anxiety (e.g., Wheaton et al., 2012; Blakey et al., 2015; Blakey & Abramowitz, 2017) . Research has also implicated disgust, intolerance of uncertainty, and anxiety sensitivity (AS)-the fear of arousal-related body sensations-as predictors of excessive COVID-19-related anxiety (McKay et al., 2020; Wheaton et al., 2021) . Although these data provide valuable information for constructing predictive models of COVID-19-related anxiety, additional psychological constructs also likely play a role.

Obsessive-compulsive (OC) symptoms, for example, remain an understudied factor in anxiety about COVID-19. Indeed, previous research has demonstrated associations between OC J o u r n a l P r e -p r o o f COVID-related Anxiety 4 symptoms and anxiety in response to prior disease outbreaks such as the swine flu (H1N1) outbreak in 2009 -2010 (Brand et al., 2013 and emerging evidence suggests that these symptoms in general are also related anxious responding to the COVID-19 pandemic (e.g., Khosravani et al., 2021b ). Yet OC symptoms are highly heterogeneous, distilling into four theme-related dimensions: (a) contamination, (b) responsibility for causing and preventing harm, (c) unacceptable thoughts related to violence, morality, sex, and religion, and (d) the need for symmetry and exactness (e.g., Abramowitz et al., 2010; McKay et al., 2004) . Few studies to date have examined the four empirically supported OC symptom dimensions as they relate to COVID-19-related anxiety. Khosravani et al. (2021a) found that all four OC symptom dimensions were associated with COVID-stress responses, including anxiety about COVID-19.

Similarly, Wheaton et al. (2021) found that all four OC symptom dimensions were associated with concerns about the spread of COVID-19, observing the strongest correlation between contamination OC symptoms and fear of COVID-19. It is therefore important to further examine how these symptom dimensions, especially contamination (e.g., Wheaton et al, 2021) , uniquely relate to COVID-19-anxiety given the highly publicized ease of transmission of COVID-19 and recommendations for frequent hand washing and use of disinfectants to prevent its spread.

In addition to these OC symptoms, certain cognitive biases are likely to predict anxiety in the context of COVID-19. The tendency to overestimate the likelihood and severity of threat, particularly dangers related to contamination, for example, has been associated with anxiety Anxiety sensitivity (AS) is another set of cognitive biases that deserves further consideration as a predictor of COVID-19-related anxiety. AS is multidimensional, involving beliefs that (a) physical sensations associated with anxious arousal are dangerous (e.g., a racing heart is a sign of a heart attack), (b) the social consequences of experiencing anxiety are severe (e.g., appearing nervous would be noticeable and lead to embarrassment), and (c) anxiety would lead to significant cognitive dysregulation (e.g., the inability to concentrate on a task equates to a serious mental problem). Of these three domains, the belief that arousal-related physical sensations are harmful may be most strongly linked to COVID-19-related anxiety since individuals with this view might misinterpret benign bodily sensations (e.g., feeling short of breath after climbing a flight of stairs) as indicating a COVID-19 infection (e.g., since it is known to affect the lungs). Indeed, emerging evidence suggests a link between AS-related physical concerns and fear of contracting COVID-19 among Chinese participants (McKay et al., 2020) .

Relatedly, individuals high in AS may be more likely to self-monitor body sensationse.g., body vigilance-in an attempt to detect and respond to threat. Body vigilance refers to a tendency to carefully monitor body sensations (Olatunji et al., 2007) and may contribute to COVID-19-related anxiety as frequent body scanning creates more opportunities to notice benign changes in one's body, which individuals high in AS then might interpret as dangerous (i.e., as signs of infection). It is important to note that while body vigilance is a more general tendency to scan one's body and not specific to monitor for signs of illness or threat, the pandemic likely activated higher levels of vigilance tilting in the threat-related direction (e.g., demonstrated that AS is associated with body vigilance (Zvolensky & Forsyth, 2002) , and that both are associated with anxiety in response to previous infectious disease outbreaks (e.g., Blakey et al., 2015; Wheaton et al., 2012) . Thus, AS-related physical concerns and body vigilance may be two additional risk factors for anxious responding to COVID-19.

With the threat of COVID-19 expected to continue throughout the next year, and with future disease outbreaks certain to occur, it is critical that we understand predictors of excessive anxiety in response to the COVID-19 pandemic. Such research is particularly vital given that dysfunctional anxiety about viral outbreaks may lead to ineffective behaviors that interfere with personal and community functioning, such as repeated hospital visits that overburden the health care system (Taylor, 2019). Accordingly, the aim of the present study was to examine the aforementioned psychological factors as predictors of COVID-19-related anxiety in an unscreened (non-clinical) sample in the United States. We hypothesized that contaminationrelated OC symptoms, overestimates of contamination-related threats, the physical concerns domain of AS, and body vigilance would each emerge as predictors of COVID-19-related anxiety.

The study included affiliates (students, faculty, staff) of a large public university in the southeastern United States. Participants were recruited through (a) a psychology department research participation pool, (b) university-wide listserv recruitment emails, and (c) a universitysponsored website that publicizes opportunities to participate in research. Responses were collected via a survey administered through Qualtrics, a secure online survey platform, between August 27, 2020 and November 5, 2020. Participants completed a demographics questionnaire J o u r n a l P r e -p r o o f COVID-related Anxiety 7 and the measures described below. Four distractor items (e.g., "please answer Never True for this item") were included among the measures to increase the probability that only valid responses from attentive participants would be included in the analyses. A total of 518 participants completed the study, participants who endorsed testing positive for COVID-19 (n = 80) were excluded from analyses resulting in a final sample of 438 participants. The study was approved by the university's Institutional Review Board, and informed consent was obtained from all participants included in the study.

Coronavirus Anxiety Scale (CAS; Lee, 2020) . The CAS is a widely used five-item measure of anxiety specifically associated with COVID-19. Items assess the frequency of anxious responding related to the novel coronavirus (e.g., "I felt paralyzed or frozen when I thought about or was exposed to information about the coronavirus") over the previous twoweek period. Items are rated on a 0 (not at all) to 4 (nearly every day over the last two weeks) scale and possible scores range from 0 to 20. The CAS has shown good internal consistency and test-retest reliability (Lee, 2020) . In their replication analysis, Lee (2020) identified a score of 5 or greater as indicating "clinically severe" COVID-19-related anxiety. In the present study, Cronbach's alpha for the CAS was .77. Anxiety Sensitivity Index-3 (ASI-3; Taylor et al., 2007) . The ASI-3 is an 18-item measure of beliefs regarding the dangerousness of anxiety along physical (e.g., "it scares me when my heart beats rapidly"), cognitive (e.g., "it scares me when I am unable to keep my mind on a task"), and social (e.g., "it scares me when I blush in front of other people") domains.

Participants rate their agreement with these items on a 0 (very little) to 4 (very much) scale.

Possible subscale scores range from 0 to 24, and total scores range from 0 to 72. The ASI-3 has a three-factor structure with good reliability and criterion-related validity in previous research (Taylor et al., 2007) . In the present study, Cronbach's alpha for the ASI-3 was .91. contamination for each item on a 0 (not at all) to 100 (extremely) scale. Separate likelihood (CCS-L) and severity (CCS-S) subscales are formed by computing the average response for items falling on the CCS-L and CCS-S subscales, respectively. The CCS total score is then created by summing the two subscales. In the present study, Cronbach's alpha for the CCS was .97. (BVS; Schmidt et al., 1997) . The BVS is a four-item measure of one's tendency to closely attend to arousal-related body sensations. The first three items assess attentional focus to, sensitivity to changes in, and amount of time devoted to monitoring body sensations on a 0 (not at all) to 10 (extremely) scale. One of these items is "I am very sensitive to changes in my internal body sensations." In the fourth item, respondents separately rate the degree of attention paid to 15 body sensations (e.g., heart rate) on a 0 (none) to 10 (extreme) scale, which are averaged to yield a single score. Possible scores range from 0 to 40, with lower scores indicating less body vigilance. The BVS has shown good internal consistency and testretest reliability in previous research (Schmidt et al., 1997) . In the present study, Cronbach's alpha for the BVS was .97.

Our approach to data analysis proceeded as follows: first, we computed mean scores on all study measures and compared across gender. Next, correlation coefficients were computed to examine relationships among all study variables. To test our hypotheses, a hierarchical linear regression analysis was performed to predict scores on the CAS. To control for potential differences based on age and gender, these variables were entered in Step 1. Next, in Step 2, the DASS depression subscale was entered to control for general distress. Then, in Step 3, the ASI subscales, DOCS subscales, CCS, and BVS were entered simultaneously. Tolerance and J o u r n a l P r e -p r o o f COVID-related Anxiety 10 variance inflation factor statistics were calculated to examine the presence of multicollinearity between predictors in the regression model.

The sample (N = 438) had a mean age of 30.29 years (SD = 17.92, range was 17 to 88).

The majority of the sample identified as female (n = 330; 75.3%), with 24.4% identifying as male (n = 107), and 0.2% identifying as gender queer or gender non-conforming (n = 1). The majority of the sample identified as White (n = 315; 71.9%), with 11.4% identifying as Asian (n = 50), 8.2% identifying as African American or Black (n = 36), 5% identifying as Hispanic or Latino (n = 22), 0.2% identifying as American Indian or Alaska Native (n = 1), 0.7% identifying as Middle Eastern (n = 3), 1.8% identifying as multiracial (n = 8), and 0.2% identifying with another racial/ethnic group (n = 1), and 0.5% of the sample missing responses to this item (n = 2). Table 1 presents the group mean scores on the CAS, DASS, ASI subscales, DOCS subscales, CCS, and BVS by gender. As expected, the mean CAS scores fell below the clinical cutoff of 5.0 (Lee, 2020) . Females had slightly higher CAS scores than did males (however this difference was no longer significant at the Bonferroni corrected alpha level). Of note, scores on the CAS ranged from 0 to 11 among males and 0 to 14 among females, indicating that some study participants had severe anxiety related to 8 .7% of the sample (n = 38), 10.3% of females (n = 34) and 3.7% of males (n = 4), reported CAS scores above the cutoff point for identifying clinically severe COVID-19-related anxiety (Lee, 2020) .

As can also be seen in Table 1 , the mean DOCS scores indicated mild OC symptoms overall. However, scores on the DOCS subscales ranged from 0 to 20, indicating that some study participants had clinically severe OC symptoms as assessed by this instrument (Abramowitz et al., 2010) . Females generally reported greater DOCS-Responsibility for Harm symptoms than males, as well as greater scores on the ASI physical subscale, the CCS, and the BVS. Table 2 presents the zero-order correlations among all study variables. As can be seen, scores on the CAS were weakly to moderately positively associated with scores on the other study measures. The CAS was most strongly associated with DOCS-Responsibility for Harm, ASI-Physical, DOCS-Contamination, and BVS. Table 3 shows the summary statistics for the final step of the linear regression model predicting scores on the CAS. Tolerance coefficients were all above 0.2 (range = .44 -.92) and

variance inflation factor coefficients all fell below 10 (range = 1.08 -2.27), indicating no reason to be concerned about multicollinearity among predictors in the regression model. In Step 1, gender and age accounted for 1.6% of the variance (F (2, 425) = 3.44, p = .03). When the DASS depression subscale was added in Step 2, an additional 6.2% of the variance was accounted for, which was statistically significant (F (3, 424) = 28.46, p < .001). Finally, addition of the clinical measures in Step 3 explained an additional 15.6% of the variance, which was also statistically significant (F (9, 415) = 9.41, p < .001). The overall model accounted for approximately 23. 

Despite increasing vaccination rates across the United States, the COVID-19 outbreak is expected to remain a significant threat to public health and safety given vaccine hesitancy and the emergence of new strains, with knowledge about the virus and recommendations continuing to evolve over the next year. Due to the extensive uncertainty surrounding the COVID-19 outbreak, it is not surprising that many Americans report experiencing anxiety about this disease (Salari et al., 2020) . When this anxiety becomes excessive, it is associated with various adverse mental health outcomes as well as maladaptive behavioral responses that have consequences for the health of communities more broadly (Asmundson & Taylor, 2020; Taylor, 2019) . Accordingly, the present study was designed to better understand the psychological factors that predict more intense COVID-19-related anxiety. The results partially supported our hypothesis in that contamination OC symptoms, AS-related physical concerns, and body vigilance each predicted more severe anxiety related to the pandemic. Yet contrary to our predictions, contamination cognitions were not predictive. Further, OC symptoms related to responsibility for causing harm emerged as a predictor, which we did not predict.

Our finding that contamination OC symptoms significantly predicted COVID-19-related anxiety was not surprising given that symptoms often involve the fear of being exposed to, contracting, or spreading infectious diseases like COVID-19. Further, given the widespread messages concerning the highly communicable nature of COVID-19 (e.g., Sun et al., 2020) , individuals with OC concerns about contamination might be especially worried about this virus.

Our findings, however, are correlational; thus, it is also plausible that OC contamination symptoms developed or worsened after COVID-19-related anxiety had emerged (e.g., Asmundson et al., 2020; Khosravani et al., 2021) . We were surprised by our finding that the J o u r n a l P r e -p r o o f COVID-related Anxiety 13 tendency to overestimate the likelihood and severity of contamination did not predict COVID-19-related anxiety. Not only did this conflict with our hypothesis and findings for contamination OC symptoms, it also conflicts with previous research suggesting a robust link between overestimates of contamination risk and excessive anxiety in response to other disease outbreaks such as the Ebola virus (e.g., Blakey et al., 2015) , swine flu (e.g., Wheaton et al., 2012) , and the Zika virus (e.g., Blakey & Abramowitz, 2017) . Of note, the scale used to assess contaminationrelated cognitions (the CCS) includes many items that measure contamination overestimates resulting from surface transmission, which is not the principal way that COVID-19 is thought to spread. This might explain our findings and represent a phenomenological difference between contamination fears in general and those specifically associated with COVID-19.

As expected, we found that AS-related physical concerns were associated with greater COVID-19-related anxiety. This is in line with previous research showing that the tendency to misinterpret benign bodily sensations as danger signs is associated with excessive concerns about health-including within the context of disease outbreaks (e.g., Blakey et al., 2015; Wheaton et al., 2012) . Thus, individuals with elevated physical AS, may be prone to misinterpreting harmless bodily sensations as evidence that they are infected with COVID-19. This in turn, may increase COVID-19-related anxiety and lead to safety-seeking behaviors (e.g., visiting multiple doctors to seek reassurance that bodily sensations or changes are not due to infection). Body vigilance, which is closely related to AS (Zvolensky & Forsyth, 2002) , also emerged as a significant predictor of COVID-19-related anxiety, as predicted. Thus, it's is possible that individuals with greater AS, who are prone to monitoring their body, are more likely to detect and misinterpret body sensations that trigger anxiety about COVID-19. In this way, AS and body vigilance may work in tandem to confer risk for excessive COVID-19-related-anxiety. OC symptoms related to responsibility for harm were also a significant individual predictor of anxiety about the COVID-19 outbreak. These OC symptoms involve distressing, repetitive thoughts of causing unwanted accidents, illnesses, injuries, or other awful events to befall oneself or others, along with checking and re-assurance-seeking rituals. Given that COVID-19 is a highly transmissible disease, it is understandable that individuals with excessive fears of this virus would endorse elevated concerns about causing harm to others as well as checking behavior. This finding is in line with recent studies that have demonstrated that all OC dimensions, including responsibility for harm, are associated with increased stress responses to COVID-19 in OCD patients (e.g., Khosravani et al., 2021a) . This finding is also notable given findings that OCD patients with responsibility for harm symptoms may be at particular risk for general distress as well as worsening OCD symptoms during the pandemic (e.g., Fontenelle et al., 2020; Nissen et al., 2020) . Fears related to causing harm to oneself or others are often associated with elevated feelings of personal responsibility. As such, it also follows that individuals with greater concerns about being responsible for harm would be more vulnerable to excessive COVID-19-anxiety. Unfortunately, as this finding was not hypothesized to be a unique predictor of COVID-19-anxiety, we did not include a measure of cognitive biases associated with inflated responsibility (such as the Obsessive Beliefs Questionnaire; Obsessive Compulsive Cognitions Working Group, 2003) in the present study. However, we recommend that future research aimed at better understanding COVID-related anxiety include such a measure given this result.

This study's findings are subject to a number of limitations. First, the cross-sectional nature of our data precludes drawing conclusions regarding causality or temporal precedence.

For instance, as discussed earlier, individuals with excessive anxiety about COVID-19 may have 

Replication analysis of the coronavirus anxiety scale

Mental health characteristics associated with dysfunctional coronavirus anxiety

A critical evaluation of obsessive-compulsive disorder subtypes: symptoms versus mechanisms

Anxiety regarding contracting COVID-19 related to interoceptive anxiety sensations: The moderating role of disgust propensity and sensitivity

Fear of the coronavirus (COVID-19): Predictors in an online study

The immediate effect of COVID-19 pandemic on children and adolescents with obsessive compulsive disorder

DOCS-C .32* .19* .23* .20*

CAS -Coronavirus Anxiety Scale, BVS -Body Vigilance Scale, CCS -Contamination Cognitions Scale, DOCS -Dimensional Obsessive-Compulsive Scale (C -Contamination, R -Responsibility for Harm, UT -Unacceptance Thoughts, S -Symmetry), ASI-3 -Anxiety Sensitivity Index -3, DASS-21 -Depression, Anxiety, and Stress Scale 21 item